IELTS Speaking Pronunciation: What the Examiner Is Actually Listening For

A quarter of your score, and the one nobody trains

Speaking is marked on four criteria, each worth 25%: Fluency and Coherence, Lexical Resource, Grammatical Range and Accuracy, and Pronunciation. Most candidates prepare for the first three. Pronunciation gets treated as fixed — something you were born with, or acquired young, or did not.

That belief costs bands, and it rests on a misunderstanding of what is being measured. The criterion is not asking whether you sound British. It is asking a much narrower and more trainable question: how much work does the listener have to do to understand you?

Your accent is not being marked

This deserves stating plainly, because the myth is persistent and it makes candidates self-conscious in a way that damages their fluency.

The Band 8 descriptor states that the candidate's first-language accent has minimal effect on intelligibility. Note what it does not say. It does not say the accent is absent, reduced, or replaced. A Band 9 speaker can have an unmistakably Indian, Nigerian, Brazilian or Polish accent. The descriptor is about effect, not origin.

What the scale actually tracks, from Band 6 upward, is the combination of two things: how wide a range of pronunciation features you use, and how consistently you control them. Band 6 uses a range with mixed control and can generally be understood, though mispronounced words or sounds reduce clarity at times. Band 8 uses a wide range, sustains flexible use with only occasional lapses, and is easy to understand throughout.

Band 7 sits between them, and it is defined in an unusual way: all the positive features of Band 6, plus some but not all of Band 8's. In practice that means an examiner hearing you needs to find real, repeated evidence of Band 8 behaviour — not one well-produced sentence, but a pattern.

The six features being assessed

"Pronunciation features" is the term the descriptors use, and it covers considerably more than individual sounds. Six things are in scope, and they are not equally important.

Individual sounds. The consonants and vowels of English. Substituting one for another matters when it destroys a meaning distinction — ship and sheep, thin and tin, live and leave. It matters much less when the substitution is systematic and predictable, because listeners adapt to a consistent accent within a minute or two.

Word stress. Which syllable in a word carries the emphasis. This one punches far above its weight, because English listeners identify words substantially by their stress pattern. Say PHOtograph, phoTOGraphy and photoGRAPHic with the stress in the wrong place and a listener may not recognise a word they know perfectly well.

Sentence stress. Which words in a sentence are emphasised. English stresses content words — nouns, main verbs, adjectives, adverbs — and reduces grammar words. Stressing everything equally is the single most common reason a technically accurate speaker sounds hard to follow.

Connected speech. How words run together: linking a final consonant to a following vowel, weakening vowels in unstressed positions, dropping and blending sounds across boundaries. An hour and a half is produced as roughly a-naw-ran-da-half by fluent speakers.

Chunking. Where you pause. Fluent speech is grouped into meaningful units with pauses between them, not between every word and not in the middle of a phrase.

Intonation. The melody — rising, falling, and the shapes that signal a question, a list, a contrast, or the end of a turn.

Of these, word stress, sentence stress and chunking are where most candidates gain the fastest ground, and all three are learnable in weeks rather than years.

Why sentence stress does the heavy lifting

English is a stress-timed language. Stressed syllables arrive at roughly regular intervals and everything between them compresses to fit. That rhythm is how listeners parse the stream into words.

Many candidates come from syllable-timed languages, where every syllable takes roughly equal time. Speaking English with that rhythm produces something perfectly pronounceable at the level of individual sounds and yet noticeably effortful to listen to — which is precisely what the descriptors penalise.

The fix is mechanical. In any sentence, decide which words carry the meaning and lean on them. Reduce the rest.

I was thinking about going to the cinema at the weekend.

The content is thinking, cinema, weekend. Everything else — I was, about going to the, at the — compresses. Read it that way and the sentence takes on English rhythm without a single sound changing.

Practise this by taking any sentence you have written for Speaking Part 2, underlining three or four content words, and saying it while tapping only on those. It feels exaggerated. It is what fluent speech does.

Chunking is a fluency feature and a pronunciation feature

Where you pause is assessed under Pronunciation, and it also shapes the examiner's impression of your Fluency and Coherence. Two criteria, one habit.

Pause between meaning groups, not inside them.

When I was younger / I used to spend most weekends / at my grandmother's house / in a small village near the coast.

Each chunk is a unit of meaning delivered in one breath group. That structure lets a listener process what you said while you produce the next piece.

Compare the pattern examiners hear constantly from Band 6 candidates: a pause after almost every word, or a pause in the middle of a phrase while the next word is retrieved — I used to spend most... weekends at my... grandmother's house. Same words, same sounds, materially harder to follow.

Hesitation while you search for language is normal and expected. Hesitation that splits a phrase in half is what damages the score. If you need thinking time, take it between chunks, and fill it with something natural — well, I suppose, let me think — rather than stopping mid-phrase.

Word stress deserves a notebook

Word stress is unpredictable in English and simply has to be learned per word. The good news is that it is finite, and the words you need in an IELTS answer are not exotic.

Some patterns that repay attention:

Suffixes fix stress. Words ending in -tion, -sion, -cian, -ity, -ical put the stress on the syllable immediately before the ending: ediUCAtion, responsiBILity, ecoNOMical. This one rule covers hundreds of academic words.

Noun and verb pairs shift. REcord the noun, reCORD the verb. Also PREsent and preSENT, INcrease and inCREASE, CONtrast and conTRAST. These appear constantly in Part 3 discussion.

Compound nouns stress the first element. CREdit card, POlice station, SWIMming pool.

Keep a running list of words you use often and mark the stress on each. When you learn a new word for Part 3 abstract topics, learn its stress at the same time — a high-level word delivered with the wrong stress often lands as no word at all, which wastes both the vocabulary and the pronunciation mark.

Intonation carries meaning you cannot express otherwise

The main thing to get right is the fall at the end of a statement. English statements fall; questions of the yes-or-no type rise. Candidates whose first language uses flatter contours often end statements level, which reads to an English listener as unfinished — and prompts the examiner to wait, or to prompt again, which eats your speaking time.

Beyond that, two patterns are worth deliberate practice because they appear in almost every Part 3 answer:

Listing. Rise on each item, fall on the last. We could improve public transport, reduce parking, and invest in cycling. The falling final item signals you have finished.

Contrast. Stress and pitch-jump the contrasted element. I don't think it's a financial problem — it's a cultural one. The jump on cultural does work that no vocabulary choice can do for you.

What to do about individual sounds

Do not attempt to overhaul your entire sound system before the test. It is slow, and it is the lowest-yield part of the criterion.

Do identify the small number of contrasts that actually break meaning for you. Common ones by background: /r/ and /l/; /b/ and /v/; the two th sounds; short /ɪ/ against long /iː/; final consonants dropped or devoiced; consonant clusters broken up with an inserted vowel.

Pick two. Work on those for a fortnight with minimal pairs — ship/sheep, very/berry, think/sink — and record yourself. Two fixed contrasts produce a noticeable gain in intelligibility. Twelve half-fixed contrasts produce none.

Also worth attention: word endings. Dropping final /s/ and /t/ affects grammar as well as pronunciation, so a swallowed plural or past tense can cost you under two criteria at once. The grammar side of that is covered in the grammar mistakes costing you band scores.

Recording yourself is not optional

You cannot hear your own pronunciation while speaking. Nobody can — you hear it partly through bone conduction, which is why recordings of your own voice sound wrong to you.

So record. Answer a Part 2 cue card, record two minutes, and then listen with one question in mind at a time. First pass: where did I pause, and were the pauses between meaning groups? Second pass: did my statements fall at the end? Third pass: pick any three long words and check the stress against a dictionary.

Three passes over one two-minute recording teaches more than an hour of unrecorded practice, because it converts a vague sense of "my pronunciation isn't great" into three specific, fixable observations.

The realistic path from 6 to 7

If you are sitting at Band 6 in Pronunciation, the descriptor says you use a range of features with mixed control and are generally understood with occasional loss of clarity. To move up you need evidence of sustained control, not perfection.

In order of return on effort:

1. Sentence stress. Lean on content words, compress the rest. Biggest single change to how easy you are to follow.
2. Chunking. Pause between meaning groups, never inside them. Also improves your Fluency score.
3. Word stress on your high-frequency vocabulary. Fix the fifty words you actually use.
4. Final falling intonation on statements, plus the list and contrast patterns.
5. Two individual sound contrasts that break meaning for you specifically.

Work on one of these at a time for a week each. Five weeks in, an examiner is hearing a different speaker — not one with a different accent, but one whose speech takes less work to follow. That is the whole criterion.

The short version

Pronunciation is 25% of Speaking and it does not measure your accent. It measures range of features and consistency of control, both in service of how easily you can be understood. Word stress, sentence stress and chunking give the fastest gains. Record yourself, listen for one thing at a time, and fix a small number of specific problems properly rather than everything vaguely.